LETTERS high blood glucose levels, and was on 2. King BM, Ranck BA, Daugherty MD, Rau Uttaradit is a province in the north- cytotoxic chemotherapy for previous CA. tertium septicemia. N ern part of Thailand. DHF is endemic Engl J Med 1963;269:467–9. 6 months. The risk for intestinal 3. Miller DL, Brazer S, Murdoch D, Reller in Uttaradit, as it is in most parts of injury is high in severe liver disease LB, Corey GR. Significance of Clostridium the country. Between 1992 and 2001, and cytotoxic chemotherapy. tertium bacteremia in neutropenic and non- three major outbreaks of DHF Intestinal mucosal compromise may neutropenic patients: review of 32 cases. occurred, in 1993, 1998, and 2001. Clin Infect Dis 2001;32:975–8. potentiate translocation of C. tertium 4. Valtonen M, Sivonen A, Elonen E. A cluster The number of DHF cases report- to systemic circulation and metastatic of seven cases of Clostridium tertium sep- ed to the Provincial Health Office foci. The second patient had no pre- ticemia in neutropenic patients. Eur J Clin from January 1992 to December 2001 disposing medical history before the Microbiol Infect Dis 1990;9:40–2. (classified by age groups) was used as 5. Kourtis AP, Weiner R, Belson K, Richards present episode that might have FO Jr. Clostridium tertium as the the estimated annual DHF incidence. resulted in acquisition of C. tertium presenting sign of a meningocele in a Case definition and categorization from the soil. Both patients had twelve-year-old child. Pediatr Infect Dis J followed the International Statistical pyrexia, , and gan- 1997;16:527–9. Classification of Diseases and Related 6. Gredlein CM, Silverman ML, Downey MS. grene of a lower limb with C. tertium Polymicrobial due to Health Problems (ICD-10). DHF cat- as the sole bacterial isolate. Neither Clostridium species: case report and egories reported in this study included patient had neutropenia when they review. Clin Infect Dis 2000;30:590–4. both DHF without shock and the were first seen. This contrasts with 7. Coleman N, Speirs G, Khan J, Broadbent V, dengue shock syndrome (the number Wight DG, Warren RE. Neutropenic entero- earlier reports of C. tertium infections colitis associated with Clostridium tertium. of cases and deaths combined). (predominantly bacteremia), which J Clin Pathol 1993;46:180–3. Dengue fever, a milder disease mani- usually occurred in patients with pre- 8. Butler T, Pitt S. Spontaneous bacterial peri- festation, was not included. existing neutropenia (3). Both patients tonitis due to Clostridium tertium. The age distribution of DHF cases Gastroenterology 1982;82:133–4. improved with or van- 9. Lew JF, Wiedermann BL, Sneed J, Campos showed that, in the 1993 epidemic, comycin and , and both J, McCullough D. Aerotolerant Clostridium children 5–9 years of age had the isolates were susceptible to these tertium brain abscess following a lawn dart largest proportion of cases, whereas in three antibiotics in vitro. Therefore, injury. J Clin Microbiol 1990;28:2127–9. 2001, the peak age of those infected we consider both isolates to be clini- was 10–14 years. The transitional cally important. The pathogenesis of Address for correspondence: Pallab Ray, stage (mean age 11.3 years) was infection caused by C. tertium is not Department of Medical Microbiology, observed in 1998. well understood, since the organism PGIMER, Chandigarh-160012, India; fax: During the observed period, the does not produce exotoxins. No evi- +91(172)744401; email: [email protected] annual mean age of DHF case- dence exists to correlate oxygen sen- patients ranged from 8.4 to 15.1 years. sitivity with bacterial enzyme produc- Despite some fluctuation, the mean tion and pathogenicity in aerotolerant age of DHF case-patients was <10 clostridia. Our report adds to the list years of age before 1996. From 1997 of recently emerging diseases caused onward, the mean age was consistent- by C. tertium. The growing accept- Dengue ly >10 years. ance of this organism as a human Hemorrhagic Fever, The incidence of DHF in children pathogen will lead to better delin- <4 years of age decreased from eation and understanding of its patho- Uttaradit, Thailand 586.0/100,000 in the 1993 epidemic genic potential. To the Editor: Dengue hemor- to 197.5/100,000 in 2001. The inci- rhagic fever (DHF) has been recog- dence in children 5–9 years of age Pallab Ray,* Anindita Das,* nized as a disease of young children in also decreased from 1,330.3/100,000 Kundan Singh,*Anil Bhansali,* the past. Three decades ago most to 676.6/100,000 in the corresponding and T.D. Yadav* reported case-patients in Thailand years. While the incidence in children *Postgraduate Institute of Medical were 3–6 years of age (1). Increasing 10–14 years of age remained Education and Research, Chandigarh, unchanged, the incidence in those India evidence shows that the age group most affected is changing (2). We 15–24 years of age increased from 122.8/100,000 to 323.5/100,000, and References report evidence that in Uttaradit, Thailand, the predominant age of from 20.0 to 52.6 per 100,000, a more 1. Henry H. An investigation of the cultural those who acquire DHF has increased than twofold increase. reactions of certain anaerobes found in Our results clearly showed that the wounds. J Pathol Bacteriol 1917;21: by at least 2 years during the 1990s. 344–85. mean age of DHF case-patients in-

1348 Emerging Infectious Diseases • Vol. 9, No. 10, October 2003 LETTERS creased from 10.0 years in the 1993 The intervening effect of vaccina- takes place and possibly by effective epidemic to 11.3 years in 1998 and to tion against Japanese encephalitis household mosquito- elimination pro- 13.2 years in 2001, as a consequence virus, a different but related fla- grams. of a decrease in the incidence among vivirus, could also explain why the children <9 years, and an increase in mean age for most cases of DHF Acknowledgments the incidence among the older age increased. Cross-reaction between We thank the staff of Uttaradit groups. This finding was similar to dengue virus and Japanese encephali- Provincial Health Office for their contri- what had been observed earlier in tis virus is well established (8). bution on dengue reports. Singapore and Indonesia (2,3). Vaccination against Japanese Some researchers have found that encephalitis virus may temporarily Jayanton Patumanond,* when the average number of annual protect persons, primarily young chil- Chamaiporn Tawichasri,* and dengue infections declines, the chance dren, against dengue infection or at Seree Nopparat† of persons acquiring dengue infections least reduce its severity, resulting in a *Chiang Mai University, Chiang Mai, declines, resulting in delays in the age decline in the observed incidence. The Thailand; and †Tha Pla Hospital, Uttaradit, when a person has experienced the cohort of these vaccine recipients Thailand first, then second, dengue infection were then exposed to dengue infec- (4). However, in Uttaradit, as well as tion later in life and exhibited diseases References in other parts of Thailand, dengue when they shifted into an older age 1. Nimmannitaya S, Halstead SB, Cohen SN, infection is endemic, with large out- group. An increase of Japanese Margiotta MR. Dengue and chikungunya breaks occurring at 2- to 3-year inter- encephalitis vaccine coverage from virus infection in man in Thailand, vals: later epidemics have also shown 96% in 1995 to 100% in 2001 (9) 1962–1964. I. Observations on hospitalized an increase in the overall incidence appeared to confirm the above expla- patients with hemorrhagic fever. Am J Trop Med Hyg 1969;18:954–71. rates. Thus, this explanation is unlike- nation. Nevertheless, areas where 2. Goh KT. Dengue—a re-emerging infec- ly to be the reason for a shift in the age Japanese encephalitis vaccination had tious disease in Singapore. Ann Acad Med distribution of DHF in Uttaradit. not been implemented also experi- Singapore 1997;26:664–70. We reviewed information that indi- enced a change in the age group with 3. Corwin AL, Larasati RP, Bangs MJ, Wuryadi S, Arjoso S, Sukri N, et al. cated that the shift in age predomi- the most DHF. A final alternative Epidemic dengue transmission in southern nance could be caused by the changes explanation is the effect of herd Sumatra, Indonesia. Trans R Soc Trop Med in places of transmission. Among immunity. Some researchers have Hyg 2001;95:257–65. these was the study in Singapore, observed that in places where dengue 4. Fischer DB, Halstead SB. Observations related to pathogenesis of dengue hemor- which proposed that an effective mos- does not occur yearly, older age rhagic fever. V. Examination of age specif- quito-control program in households groups have higher rates of infection ic sequential infection rates using a mathe- had resulted in changes in which age (10). However, dengue cases had been matical model. Yale J Biol Med group had the largest number of DHF reported every year in Uttaradit, and 1970;42:329–49. 5. Ooi EE, Hart TJ, Tan HC, Chan SH. cases (5). A significant (p<0.001) rise the intervals between each epidemic Dengue seroepidemiology in Singapore. in seroconversion in children >6 years were not long. We therefore, believed Lancet 2001;357:685–6. of age coincided with the start of for- that the herd immunity hypothesis did 6. Strickman D, Sithiprasasna R, Kittayapong mal schooling. The likelihood of not explain the observed changing age P, Innis B. Distribution of dengue and Japanese encephalitis among children in dengue infection increased with time predominance in our study. rural and suburban Thai villages. Am J Trop spent away from home, suggesting The mean age of DHF case- Med Hyg 2000;63:27–35. that the location where dengue was patients in Uttaradit, Thailand, 7. Pongsumpun P, Tang IM. A realistic age acquired may have changed (5). The increased by >3 years between 1992 structured transmission model for dengue hemorrhagic fever in Thailand. Southeast recent study in Thailand also suggest- and 2001. This phenomenon may be Asian J Trop Med Public Health ed that, although dengue infection important from a public health stand- 2001;32:336–40. may be transmitted in the home envi- point, as community and health-relat- 8. Chang KJ. Studies on the serological cross- ronment, transmission within schools ed personnel may still perceive DHF reaction between dengue and Japanese encephalitis. J Microbiol Immunol Infect may also be important (6). as a disease of only small children and 1997;30:207–18. The changing of the population unintentionally leave older children 9. Uttaradit Provincial Health Office. Annual age structure also explained the age less protected or ignored. Further report 2001. Thailand: Communicable shifting phenomenon in some studies study is needed to confirm that the age Disease Control Unit, Uttaradit Provincial Health Office;, Thailand; 2002. (7). In Uttaradit, however, changes in group shifting of DHF predominance 10. Kaplan JE, Eliason DA, Moore M, Sather the age structure of the population can be explained by the changes in GE, Schonberger LB, Cabrera-Coello L, et were small from 1992 to 2001. locations where disease transmission al. Epidemiologic investigations of dengue

Emerging Infectious Diseases • Vol. 9, No. 10, October 2003 1349 LETTERS

infection in Mexico, 1980. Am J Epidemiol ant S. Typhimurium (1). In this letter, Antimicrobial 1983;117:335–43. the Table is based on the original table. However, two additional com- Drug–resistant Address for correspondence: Jayanton parisons have been added: the p val- Salmonella Patumanond, Department of Community ues, which are not based on the data Typhimurium Medicine, Faculty of Medicine, Chiang Mai but are approximations based on the University, Chiang Mai 50200, Thailand; fax: parameters in the table. (Reply to Dahl) 66+53+945476; email: [email protected] The conclusion is that only In Reply to Dahl: The emergence quinolone resistance is associated and spread of multidrug-resistant with excess mortality compared with Salmonella enterica serovar nonresistant isolates. Penta-resistant, Typhimurium DT104 (MDR DT104) quinolone-susceptible S. Typhim- contributed to an international urium has a risk ratio of 2.9 (1.1 to increase in antimicrobial drug resist- Antimicrobial 7.9) compared to the ratio of ance in S. Typhimurium in the late Drug–resistant non–penta-resistant isolates 2.1 (1.5 1990s (1,2). This type of Salmonella to 2.9). When these figures are com- Salmonella is usually resistant to five drugs: pared, the approximate p value is ampicillin, chloramphenicol, strepto- Typhimurium 0.55, which, of course, is far from mycin, sulfonamides, and tetracycline (Reply to Helms) being significant. Thus, on the basis (R-type ACSSuT) and easily acquires of the article by Helms, penta resist- resistance to other drugs, including In Reply to Helms: In the article ance may not pose a greater threat to quinolones, trimethoprim, and amino- by Helms et al., Helms concludes that human health than non–penta resist- glycosides (1,3–5). To determine infections with Salmonella Typhim- ance. However, the measured effect of death rates after infection with MDR urium strains resistant to ampicillin, penta resistance is achieved by the DT104 or closely related strains, we chloramphenicol, streptomycin, sul- inclusion of quinolone-resistant S. identified patients who were infected fonamide, and tetracycline (hereafter Typhimurium in the group. with strains at least resistant to referred to as penta-resistant) were ACSSuT (6). Analysis limited to associated with higher death rates than Jan Dahl* strains that were only R-type ACSSuT infections with non–penta-resistant S. *Danish Bacon and Meat Council, would have given a misleading result Copenhagen, Denmark Typhimurium. Helms also concluded since MDR DT104 often, as men- that infections with quinolone-resist- tioned, develops additional resistance Reference ant (nalidixine-resistant) S. Typhimu- to other classes of antimicrobial drugs rium were associated with higher 1. Helms M, Vastrup P, Gerner-Smidt P, in addition to the ACSSuT-complex. death rates than quinolone-susceptible Mølbak K. Excess mortality associated This fact needs to be taken into S. Typhimurium (1). with antimicrobial drug-resistant Salmonella Typhimurium. Emerg Infect account in any attempt to quantify the Table 2 in Helms’ article provides Dis 2002;8:490–5. overall public health impact of MDR information that enables close scruti- DT104 and related strains. ny of this conclusion and comparison Address for correspondence: Jan Dahl, Danish We found, in our matched cohort of the excess mortality associated Bacon and Meat Council, Axeltory 3, 1609 study (6), that 283 patients infected with penta-resistant, quinolone-sus- Copenhagen V, Denmark; fax: 4533145756; with strains resistant to at least ceptible S. Typhimurium with the email: [email protected] ACSSuT were 4.8 times more likely excess mortality of non–penta-resist- to die than the general Danish popula- Table. Table showing additional comparisons (1)a tion, compared with 2.3 for 953 Resistant Susceptible patients infected with pansusceptible Deaths/ Deaths/ strains. This difference in death rates cases RRb (95% CI) cases RRb (95 % CI) p value Penta with and 12/283 4.8 (2.2 to 10.5) 47/1,764 2.1 (1.5 to 2.9) 0.06 ocurred mainly because 40 of the 283 without quinolone strains had R-type ACSSuTNx (i.e., Penta with 5/40 13.1 (3.3 to 51.9) 47/1,764 2.1 (1.5 to 2.9)c 0.01d additional resistance to nalidixic quinolone acid), and infection with this strain in Penta without 7/243 2.9 (1.1 to 7.9) 47/1764 2.1 (1.5 to 2.9)c 0.55cd quinolone particular is associated with a high aRR, relative risk; CI, confidence interval. death rate (relative mortality 13.1). As bAdjusted for coexisting conditions. cCompared to the non-penta group. Dahl suggests, infection with R-type dApproximations based on the parameters from the table. ACSSuT (Nx susceptible) was not

1350 Emerging Infectious Diseases • Vol. 9, No. 10, October 2003